Walking and running are behaviors that can be maintained for many years. Our national cohort of 45,704 walkers reported walking >12 mi/wk for an average (+_SD) of 9.1+_10.1 years. Our cohort of 120,000 runners ran >12 mi/wk for an average of 12.4+_7.6 years. Yet a significant portion of runners (and presumably walkers) reduces their exercise each year (41% reduced their mileage >50% during 7.6 yrs of follow-up). The risk for exercising less appears to decrease with the number of years run, race participation, and better race performance. We hypothesize that an intervention that promotes participation in walking and running events, provides social support, and engenders intellectual interest and a sense of accomplishment will serve to maintain these behaviors. Moreover, we hypothesize that the Internet can provide this intervention to large populations at nominal cost. We propose to test this hypothesis in a 3-year randomized trial of 8,000 walkers and 8,000 runners. Subjects having Internet access will be recruited from our existing cohorts of over 160,000 walkers and runners. The treatment group will consist of 4,000 runners and 4,000 walkers who will be referred to our interactive Web site. This site will build upon our current Internet-based survey tool that provides full-function physical activity and diet logs (www.healthsurvey.org). We will add the capacity to: 1) track daily mileage on a virtual cross-continental walk or run from Yorktown Virginia to Florence, Oregon and to view cross-continental progress (http//trailmaps.lbl.gov); 2) choose anonymous running and walking partners with similar goals to travel with along the virtual course; 3) communicate with partners to provide mutual support; 4) provide recognition for completing each state on an awards page; 5) download video clips of regional 10km and marathon races to entice participation (these will enable participants to visualize participation and completion of the races); 6) provide information on regional races, walking and running clubs, and mentoring services. The control group will consist of 4,000 runners and 4,000 walkers who will be referred to Road Runners of America (runners) or Volkssport Websites (walkers). Both the treatment and control groups will receive by mail baseline and follow-up surveys on physical activities. A subset of 100 runners and 100 walkers will be given pedometers with heart rate monitors to verify activity. Behavioral models will be developed by comparing walking and running activity to: 1) Web site activity; 2) additional data solicited periodically and at times of acute change in exercise level using E-mail. We estimate that we will be able to detect differences in exercise cessation rates between the intervention and control group of 0.45% in runners and walkers separately (alpha=0.05, 80% statistical power).